Leaderboards · Hardware (stock)
Hardware at stock.
3DMark and Cinebench numbers on consumer GPUs and CPUs in their out-of-box configuration. Modded entries (repad, repaste, undervolt, fan curves) live on a separate board so a mod-aided score never outranks a stock one.
Scoring methodology: v1
See also: modded entries (separate board).
3DMark
Multiple variants per test cover BIOS / power-limit / Windows-power-plan sweeps on the same hardware. Δ vs own-historical compares against the owner's prior 2024 runs on the identical card; Δ vs public-typical compares against the public reference range.
| Test | Variant | Score | Gfx | CPU | Δ own | Δ typical | Pwr peak | Hot peak |
|---|---|---|---|---|---|---|---|---|
| Time Spy | TimeSpy | 26547 | 28030 | 20425 | -7.2% | -4.5% | 320 W | 102.8 °C |
| Time Spy | TimeSpy_clean | 26294 | 28146 | 19154 | -8.1% | — | — | — |
| Steel Nomad | SteelNomad | 6632 | 6632 | — | -3.2% | — | — | — |
| Time Spy | TimeSpy_220W_UNINTENDED_round2_bios2107 | 26948 | 28510 | 20565 | — | — | — | — |
| Time Spy | TimeSpy_210W_round2_bios2107 | 26758 | 28334 | 20349 | — | — | 193 W | 79 °C |
| Steel Nomad | SteelNomad_220W_UNINTENDED_round2_bios2107 | 6630 | 6630 | — | — | — | — | — |
| Steel Nomad | SteelNomad_210W_round2_bios2107 | 6600 | 6600 | — | — | — | 148 W | 78 °C |
| Speed Way | speedway_run | 6652 | 6652 | — | -0.4% | — | — | — |
| Speed Way | speedway_round2_210W_bios2107 | 6721 | 6721 | — | — | — | 183 W | 72 °C |
| Time Spy | TimeSpy_210W_HighPerf_checkpointA_step1 | 26696 | 28299 | 20212 | — | — | 205 W | 80 °C |
Cinebench R23
| Test | Variant | Score | Δ typical | Pwr peak | Pkg peak |
|---|---|---|---|---|---|
| Cinebench R23 CPU Single Core | round1-bios0404 | 1927 | -6.7% | 118 W | 74 °C |
| Cinebench R23 CPU Single Core | round2-bios2107 | 1942 | -5.3% | — | 74 °C |
Cross-card
Same workload, two cards, same workload window. Both cards in stock configuration; the 3090 numbers were captured before the pad swap (see /mods).
| Test | Card A | A score | Card B | B score | Δ A→B | Note |
|---|---|---|---|---|---|---|
| Steel Nomad | RTX 3090 (stock) | 5026 | RTX 4080 SUPER | 6600 | -23.8% | Raster gen gap (Ampere → Ada) |
| Speed Way | RTX 3090 (stock) | 5411 | RTX 4080 SUPER | 6721 | -19.5% | RT closer; Ampere RT cores hold up |
| Time Spy (overall) | RTX 3090 (stock) | 18747 | RTX 4080 SUPER | 26758 | -29.9% | Wider; CPU subscore drag (12900KF vs 13700K) |
| Time Spy (Graphics) | RTX 3090 (stock) | 19502 | RTX 4080 SUPER | 28334 | -31.2% | Pure GPU delta — high end of normal Ada-vs-Ampere |
Raw data
Methodology: /leaderboards/methodology. JSON API: /leaderboards/hardware.json.